filter

Apply binary keep-or-drop criteria to item lists while preserving source attribution.

Updated Feb 20, 2026

Search Tags:#provenance #pipeline #deduplication #filter #prune #keep #drop

Ready to run

npx skills add https://github.com/supertyrelle/pelley --skill filter-supertyrelle

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: filter
Source: https://github.com/supertyrelle/pelley/tree/main/skills/filter
Command: npx skills add https://github.com/supertyrelle/pelley --skill filter-supertyrelle

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It removes irrelevant items from a list by applying a clear yes-or-no criterion, so you can prune noisy results after gathering, distilling, or ranking.

Core Features & Use Cases

Binary keep/drop decisions: Evaluate each item against a criterion and either keep it or remove it.
Pipeline-aware pruning: Work from prior primitive outputs while preserving source, confidence, and detail for the items that remain.
Flexible filtering: Support positive rules like confirmed only and negative rules like not stale or not duplicate.
Use case: Clean up a research backlog, remove duplicates, or exclude low-confidence findings before the next analysis step.

Quick Start

Use the filter skill to keep only confirmed security-related findings from the current list.

Frequently Asked Questions about filter

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I remove duplicates and prune noisy items from a list?▼

To prune noisy items and remove duplicates from a list, you apply a binary keep-or-drop criterion against each item. This filtering process evaluates conversational items or pipeline outputs, dropping irrelevant data while preserving provenance and confidence details for the remaining items.

What is the best way to filter low-confidence findings before analysis?▼

Filtering low-confidence findings before analysis requires applying a binary criterion to exclude items below your confidence threshold. This approach conservatively drops unconfirmed results from your research backlog while retaining source attribution and confidence scores for the surviving items.

Can I apply positive and negative filters to pipeline outputs?▼

Yes, you can apply both positive and negative filters to pipeline outputs. Positive filters keep only confirmed items matching specific criteria, while negative filters exclude stale or duplicate data, ensuring pipeline-aware pruning removes irrelevant results without losing source attribution.

Does filtering preserve source attribution and provenance for kept items?▼

Filtering preserves source attribution, confidence, and provenance for kept items. When applying binary keep-or-drop decisions to prior primitive outputs, the process removes non-matching items entirely while maintaining the original metadata and source details for the remaining filtered set.

When do I need to use a binary keep-or-drop criterion for deduplication?▼

You need a binary keep-or-drop criterion for deduplication when cleaning up a research backlog or excluding redundant items before the next analysis step. This approach evaluates each item against a clear yes-or-no rule to remove duplicates and noisy results.

What are the limitations of using conservative default behavior for list pruning?▼

The limitation of using conservative default behavior for list pruning is that it may retain ambiguous items that a stricter filter would remove. This cautious approach ensures you do not accidentally drop potentially relevant data, but it might leave some noise in your filtered results.